Fog computing addresses the issues experienced in Cloud computing such as latency, delay, and bandwidth while provisioning the resources such as computation, storage, and network for IoT applications. Fog resources and their integration with Cloud poses other significant problems such as data security, overhead in data acquisition, data tampering, data privacy, data breach, etc. Designing a secure, resilient, and scalable infrastructure for IoT network is challenging because of resource heterogeneity, resource-constrained devices, and unreliable network connectivity. The researchers have developed many chaos-based encryption algorithms to enhance the security level and make it difficult for intruders to access the data. However, such traditional algorithms increase the communication overhead and computational cost. This necessitates the design of new secure communication protocols that comply with various security requirements of Class of Fog Applications (CoFA). This work proposes an efficient and intelligent method that classifies the applications arriving at the system into CoFA and maps them to five levels of chaos-based security protocols for secure communication. An adaptive sliding mode security schemes are developed using Genesio-Tesi, Modified-Jerk, Chua's chaotic, and Financial systems in CoFA framework for secure communication. A comparison-based performance study, with state-of-art, reveals the efficacy of the proposed method, which performs better for security optimization and synchronization with the least error.